6.30.25 Weekly SMB Agent AI newsletter

Hello All,

This is my second weekly newsletter on AI for small / medium size companies, going out to those we work with at Naibu / have worked with, other connections in the industry, collaborators etc. We're 487 subscribers, with 23 new subs last week. The whole newsletter is very much early / WIP but fun and hopefully useful.

Our overall these is that companies should think of AI as intelligence to be deployed and thoughtfully integrated with their teams, not just software. That mindset changes how you approach this new era and the opportunities it offers. It is already starting to take on much of the operational work and this will only grow over time.

The goal here is to track the signal and highlight the noise or distractions and try navigate the space more fruitfully.

Signal versus Noise section - Is AI impacting business productivity?

We know businesses are investing in AI. Is it hype or a deep durable structural change?

Weekly Signal / Noise:

• Signal📡:
  • AI is already impacting entry level jobs and from our experience we can see why. It is just really good at basic operational work but it still needs guardrails.
    • "AI cut hiring of new grads by 25% at tech giants like Google, Meta" - AI disrupting entry-level job market for college graduates: Report
    • "AI is reported to be writing around 50% of some tech firms’ code, and one AI boss said half of entry-level white-collar jobs could soon be lost" - AI jobs backlash
  • Guardrails example: Here is a perfect example of how AI still needs a lot of help.
    • "The $1000 Test That Breaks Every AI Model Out There Today" - Claude failure as vending machine manager
    • "Anthropic’s Claude AI became a terrible business owner in experiment that got ‘weird’" - Tech crunch review of same experiment.
• Noise💥:
  • OpenAI "Enterprise" / Teams (with connectors) and Claude MCPs. This is kind of my bugbear at the moment. Companies selling the idea that you can usefully leverage the foundational AI solutions completely thru their own solutions. It might work for consumer but for companies it is dead end.
    • Having reviewed both of options, these are again, at best, personal productivity tools, not company wide solutions.
    • The primary issues are:
      • You have to use their web UI for all interactions, rather than your company systems.
      • They can't properly access your company systems
      • Their configs have some serious limitations impacting even basic tasks.
      • The real upside is to the large foundational companies as it helps them:
        • Lock you into their ecosystems
        • Learn about you and your habits, so they can sell more products later.
      • Practical examples:
        • Slack deep dive:
          • In this instance the solution is really just AI on top of the data dump of slack messages. So it can summarize what's been said etc but what you want is the AI to be INSIDE slack itself, like a team member.
          • It can't take any actions beyond summaries, message someone etc.
        • Email example:
          • While they can connect to your email server (at least Gmail and Outlook), they are limited by:
            • The fact that they can't review emails in real time, they need you to prompt them do so. The real value proposition here should be that the AI responds in real time, with your approval, not that you have to watch your emails to ask it to respond.
            • They can't access your other companies systems in coordination with their actions to save attachments or the email itself to a system, attach files to emails, look more deeply at the relationship with the recipient / their company etc.
            • Their ability to only access a certain limited number of emails at a time.

Comments and links for the week:

• Context engineering. This topic is taking over prompt engineering as the new frontier of best practices.
  • A couple of good references:
    • The AI Breakdown podcast on CE
    • Nate Jones overview of CE
• Competition for OpenAI.Anthropic ARR of $4B versus OpenAI of $10B. I thought the gap was wider.

People / teams I follow constantly for insight:

• Andrej Karpathy - Technical genius
• Nate Jones - Product manager / AI Agent deep diver
• All-in pod - Can be nauseating but they are smart and connected
• BG2 pod - Like All-in but less nauseating
• John Vervaeke - The most important public intellectual on matters of the mind, cognition, meaning and what it means to be human.
• Bernardo Kastrup - Computer Scientist and Philosopher
• More to come...

Ciao for now. Have a great July 4th!

Check out what we do at Naibu.org or the blog.